Kyle Geers, CEO and Co-founder of Zeroed In Consulting, joins Adam and Tom to discuss the role of AI in accounting. They delve into how AI can streamline processes, provide valuable insights, and improve efficiency in the accounting industry. Kyle shares his experiences of incorporating AI into his consulting firm's services and the potential benefits and risks involved. The conversation also touches on the importance of human review in AI processes and the importance for accountants to understand AI.
“I don't necessarily think that AI is going to replace accountants, but the accountants who use AI are going to really succeed.” – Kyle Geers
The finer details of this episode:
Episode resources:
Intro (00:00:00) - Welcome to the modern CPA Success Show, the podcast dedicated to helping accounting firms stay ahead of the curve. Our mission is to provide you with the latest and greatest insights on cutting edge tools, innovative marketing strategies, virtual CFO services, and alternative billing methods. Join us as we change the way people think about accounting.
Tom (00:00:22) - Adam, we are talking about artificial intelligence today with Kyle Geers. Kind of initial thoughts. What are people going to hear?
Adam (00:00:31) - It's scary right? Like AI is going to take over. We might as well just kind of figure out our exit strategy now. No. I think AI is going to really help us. I mean, I think, what Kyle talks a lot about is just being able to, utilize it and leverage it and be on the front end of it. And I think as long as we're doing that and we're constantly exploring the new ideas and the things that it can bring to the table, I think we'll be able to, you know, leverage it to improve our lives and make us more valuable to our clients.
Tom (00:01:01) - So I think you're totally right. The couple of things I would add to what you had said is he's a practitioner. He's been doing accounting, both big firm and started his own. So he's actually using it. He's not just out kind of a proponent of AI in the point I can't remember if he's either you or Kyle was AI's already in a lot of the things that we're doing, and so we are experiencing it. And so the idea of sort of using it for the way you want it to versus just having apps that have AI in it, I thought was an important piece. So I think people are going to really get a lot out of this episode.
Adam (00:01:33) - Yeah. No, I think so too.
Tom (00:01:36) - All right. Welcome to this episode of the modern CPA Success Show. I'm Tom Wadelton. I'm co-hosting with Adam Hale. Adam. Welcome today.
Adam (00:01:43) - Hey, Tom!
Tom (00:01:44) - We are excited to talk to Kyle Geers. And Kyle is CEO and co-founder of Zeroed in consulting. Kyle welcome and as form of intro you wanna just tell us a little bit about kind of your career journey and what got you two to today's conversation?
Kyle (00:01:58) - Yeah, absolutely. Tom and Adam, thanks for having me on. Glad to be a part of the show. So, yeah, just a quick background, for myself. So, started probably where a lot of CPAs start. You know, I worked at a large global CPA firm of Grant Thornton. So I'm based here in Los Angeles. And so, started on the audit side, right, of, you know, working with, helping out companies do their financial statement audits, you know, public integrated audits, really spans the range of, you know, industries going from startups to public companies. It was a good experience to get a lot of different, see a lot of different sides of public accounting and see what works for companies and what doesn't.
Kyle (00:02:39) - Right. And, so I stuck I was I've been in the public accounting industry for a while. It's been probably about 5 or 6 years on the audit side and eventually made the decision to, transition over into the consulting side. So, you know, now helping companies get ready for those audits. Right. And so, you know, they were starting, the firm was starting a technical accounting advisory group at the time, relatively young. And so I was kind of one of the local leaders out there on the West Coast to, you know, really help build the business and build the group. And so, you know, that was probably about 3 or 4 years there and, you know, got a lot of experience of, you know, helping go out and win the work, right, or build the team or do some recruiting. And so came to find that I really enjoy, you know, I'm kind of one of those, I call it nerdy technical accountants that really enjoys the accounting.
Kyle (00:03:30) - But, you know, I also really enjoyed the business building side of it. So, eventually connected with my, with my partner, my co-founder, who, you know, he was at Grant Thornton as well, and took a different path for a while, but we reconnected and we said, hey, why don't we do this for ourselves? Right. And there's some ideas that we have in our head that we would do differently, you know, than a lot of other public accounting firms out there. So starting in February 2021, we made the choice to start zero in consulting. And, you know, really what we do, our bread and butter was right out the gate was, you know what we knew how to do, right? So we do a lot of, helping companies with their technical accounting, you know, US gap, complex accounting, services like that, and you know, where my partner's, focus is, is on the inside, right? So helping out in a controller ship capacity, you know, helping with month and close.
Kyle (00:04:22) - So we kind of focused on more of those complex accounting tasks. but one of the things that, you know, right out the gate that we wanted to make sure was embedded in, you know, the culture and everything we do is, you know, incorporating technology. Right? There's so there's so much out there on the horizon, having been released in the last few years that, you know, you can use to make your processes way more efficient, way more accurate. And so, you know, maybe like 5 or 10 years ago, that was just the fortune 500 companies investing billions of dollars into technology. That's not the case anymore, right? There's things that you can do even with a simple, Microsoft Office subscription. So that's kind of our goal is to really work with, growing companies, businesses and, you know, bring that power of the technology to them. So to help them. So, you know, they don't you don't necessarily need to be a massive company to enjoy a lot of this, a lot of this automation and a lot of these efficiencies.
Adam (00:05:20) - Yeah. Kyle, you're so 2021, huh? So I feel like sometimes you get an advantage just starting out. one, you know, you're a little bit more nimble, but I kind of, you know, it's like, it's almost like software platforms. So you think about, like, QuickBooks desktop and then and then moving to just getting used to QuickBooks online. You know, you have this like advantage where you can do all these cool things. It's like I remember whenever we started our practice, like nobody was paperless and nobody was like leveraging, some of the like go to meeting at that time, technology. So we were like, oh, we're cutting edge. We can do all this, like cool stuff. And, you know, it wasn't like we had to rebuild the wheel with, you know, 200 team members so we could still be nimble and we could still try out every software. I mean, don't get me wrong. Like, we would exhaust our small team at that point, like, because we were constantly switching hands and doing stuff.
Adam (00:06:10) - But that kind of brings us into this topic. It feels like you're like starting out within the cusp of all this AI stuff. So you really have an opportunity, like you said, to embed all of that into your baseline offering. And you don't have to, like, go back and revisit all your processes you've built over the years, all the tools that you've used and figure out how to integrate it in. You can actually just kind of build it from scratch, you know, with some of these tools, obviously they'll all super change. Like we were joking earlier, but, you know, super change in a couple of years. but anyway, I just think it's a cool opportunity and love to hear, you know, how you're leveraging AI in how the rest of us can start to kind of think about that stuff? Both in, you know, both internally for our practice and then also, for our clients as well.
Kyle (00:06:58) - Yeah, absolutely.
Kyle (00:06:59) - And you made a good point, right? We got pretty lucky, I think, of, you know, being at this, this point in time where, you know, it did become widely available. Right. And so, you know, being one of the folks out there that was, you know, said that, hey, there's a wave coming, right? Like, let's we can still be on top of that wave, you know, and, let's see if we can make that happen. And like, you said, it's a lot easier to build from scratch than, you know, refining existing processes. Not that it's not that it's impossible, but it's definitely helped us so far.
Tom (00:07:30) - Yeah. So I'm curious, where do you think the biggest opportunities are if someone's looking at AI and accounting, especially the kind of practice that you're doing, where are the biggest opportunities for gains there?
Kyle (00:07:40) - Yeah. So there's a few that come to my mind right out the gate.
Kyle (00:07:43) - Right. And I think for a lot of, you know, firms, you know, a lot of companies run on pretty the same softwares. Right. So we're thinking about QuickBooks online. You know, if you're taking the next step up, you're going to a NetSuite or a sage in tact or other software providers like that. Right. And so, it's really cool to start seeing the features that are being released for a lot of these, these providers, you know, for instance, things like QuickBooks, right? They're big push over the next year or two is to, you know, introduce QuickBooks AI. Right. And they've already started trying to do that with some insights. you know, on the on the software as it stands. So, you know, it's I'm really interested to see because, you know, we're getting to this point where you can, you know, bring in something like an invoice, an AP invoice, and that can, you know, go into an inbox and that can get recognized and record it.
Kyle (00:08:31) - Right. They've already started putting rules in there that'll just automatically, categorize a transaction. You know, it's still up and coming. Right? You still definitely need to be looking at, you know, transactions like that. But, you know, as it gets refined, as it learns, I could see that really changing things for a lot of accounting firms out there where, you know, all of a sudden, you're not spending, you know, 10 to 20 hours sorting transactions, making sure you get it right. You know, you're more of doing some review, but, you know, then you're kind of able to review and analyze, you know, what does this mean, right? Doing more of that strategic side of things?
Tom (00:09:10) - Yes. Around like data entry bookkeeping. And we've seen some of those things like build Com and Expensify and DV all have some of those kind of things. And you're right, it is really good how well it can go through. It's not perfect but it takes a really good first step to try to apply some of those rules.
Adam (00:09:27) - The, the poor accounting industry, they finally get caught up with something and then we just take it away. Like I feel like Cass 1.0, everybody just finally started like going, oh, accounting firms should be in charge of all the accounting integrity and outsourcing your finance department. And it's like, oh, but not this part anymore, you know? Ta da! You're here in a year or two, QuickBooks will do it automatically and so will there every other software. So time to move upstream to advisory. So the learning curve is pretty steep there. But there's some really cool tools, like you said, for that as well.
Intro (00:09:59) - Elevate your firm with a virtual CFO playbook, how to land $60,000 per year clients and provide a killer client experience. This comprehensive online series of modules provide you with the essential tools to create and deliver scalable VCFO services developed from summit's successful approach, which propelled them from $500,000.
Intro (00:10:20) - To over $10 million in revenue in ten years. This playbook will equip you with the skills to take your firm to the next level. Additionally, upon completion, you can receive 24 hours of CPE credit for the course. Ready to grow your firm? Enroll now at VCFO.playbookSummitCPA.net.
Adam (00:10:42) - Yeah. So what are some of the things that you are deploying right now, just or even in an early infancy, you know, what are some of the things that you're doing right now?
Kyle (00:10:52) - Yeah. So, you know, on our end, we're doing a lot of the more technical work, right. And so trying to figure out how do you do, how do you incorporate, you know, AI into some of your, more complex areas. Right. And I it brings me back to a conversation that, that we had at my old firm, old group where I was in all hands call and it was you know, someone asked, right? So is technology going to replace accountants? Right.
Kyle (00:11:17) - The age old question that we've been asking for a while. and, you know, he said I could see it happening for, you know, some of the simpler accounting tasks for, you know, even audit things like that. And, he said, but it won't happen to us. We're too complex, we're too technical, we're too subjective. Right. And it was like even at that time, this was probably, you know, back in 2019 or so, maybe 2018. you know, that just immediately kind of, you know, set up a red flag in my head of, okay, there are a lot and there's a lot of people like this that are thinking the same way. Right? Like I'm protected. I'm too technical, I'm too subjective. It's not going to happen. and then, you know, fast forward to now, you know, we have these crazy, you know, AI chatbot capabilities, things like ChatGPT that if you know how to use it, if you have the right use cases, you know, you can use it in that capacity.
Kyle (00:12:09) - So, you know, to your point, Adam of, you know, how are we trying to use it? There's a lot of information out there, you know, especially on the technical side, you got to do a lot of research on, you know, thousands of pages, guides. How does this apply to me? you know, it's a really good data, data gathering, resource to say, hey, I have a question, you know, can you bring back some results? Can you bring me back a summary of how the accounting would go for this? Right. How would you account for a lease, how you would account for, you know, new debt or something like that? And, you know, I think that that's going to be a huge time saver. It's we've already started using it ourselves of gathering that research, and just having a good starting point versus, you know, having to go through hours of research, of finding where, where and what you need to look at.
Tom (00:12:59) - And to what? I'd love to dig into that a little bit. To what extent do you trust those resources, and what are the main ones that you would go after? If you're saying start your research using an AI engine?
Kyle (00:13:09) - Yeah, it's gotten a lot better than it was, you know, a year ago. Right? Like if you just kind of popped a really complex accounting question into, you know, the first iteration of ChatGPT, well, that's not going to be great. Right? And there's still and it's a good point to mention of like, you know, there's still a lot of risks, there's still a lot of attention that you need to be, you know, taking with the results that come back. Right. There's a lot of hallucinations still happening within these, you know, these chat bots. So finding, you know, it's it's validate, validate, validate. Right. It's making sure that what I see here, you know, maybe it can point me to the resource that it came from.
Kyle (00:13:48) - Right. Did it come straight from the accounting codification. You know. And if so then great. You know that's a great source. Did it come from, a firm guide? Right. You know some maybe something from KPMG that says, you know, this is how we would think about trading this. That's awesome. You know, if you can point and go and validate where that is coming from, then, you know, that's ideal. If it's coming from, you know, a lot of different sources, it might not even be right. And so you need to be double checking and making sure that that is the case, especially with these more technical areas. you know, but even to that point of getting to that starting of, okay, here's the information that's provided to me going and trying to find it or validate it. You know, that's a lot quicker than going leafing through all the research guides and getting your own answer. Initial.
Tom (00:14:41) - Yeah, I can especially see people that may be smaller firms where there's not a where there's not the guru sitting somewhere in the ivory tower, like, hey, go to them.
Tom (00:14:48) - They have the answer. To be able to get that kind of start is huge. Yeah, yeah. Adam mentioned sorry. Go ahead.
Kyle (00:14:54) - And you know, the last thing I'll say is, you know, just with the volume of, you know, with the wide range of, you know, very general information over the past year or so, it's going to be less effective. Right. But as you are able to refine it into very more like specialized data. Right. So if I'm just looking at the the accounting guidance, if I'm just looking at these guides, you're going to get a lot higher quality information. And so, you know, I don't know if you both saw the, the recent, you know, notification from, from OpenAI saying, hey, you can create your own custom GPT now. And, you know, so basically a firm could take their own instance. They could say, hey, here are the files that I want to put into this. Or, you know, here are the websites that I want to incorporate into this.
Kyle (00:15:42) - And now you're you're filtering out a lot of that kind of lower quality information out there, and you're able to really refine that a little bit. And so that's where I'm getting excited of, you know, seeing if can you do that. Can you, you know, make sure that you're allowed to, you know, incorporate the files that you are getting in there. but you know, that that's going to give you a lot higher quality results when you're asking a question.
Adam (00:16:07) - Yeah. Yeah, that's kind of where our heads been. I know we, you know, our IT team's been thinking through how to leverage AI in. Tom, I'll have you probably talk more about the one that Jake and Jamie came up with in terms of like, the coding stuff, but the one that's top of mind for us right now is actually internal. And to your point, it's less about research. You know, we definitely don't want to be I don't know, maybe it's a wise tail now that law firm that like where they made up all those court cases or whatever and made it sound very professional and they use it or whatever, but we definitely don't want to be them.
Adam (00:16:38) - And so, yes, validate sources, do all that kind of stuff. And don't get me wrong, I think that a lot of people in our team are starting to use ChatGPT for that first line of defense, or how to write stuff, or how to proofread their stuff. but what we're thinking about AI for is more of like an internal wiki. So we have a lot of really good internal conversations about topics. We, bring in a lot of subject matter experts that are good at different things. and then we also have client conversations about some of those topics, you know, like when we're explaining what a phantom plan is or, you know, doing some of these other advisory, related topics. So we're looking at is there a way to use a, you know, kind of a note taking software, you know, similar to built in to zoom, but there's like 20 or 30 of them out there. I think we've kind of landed on the one that we're going to ultimately end up using.
Adam (00:17:29) - But be able to bring all that information in, let it automatically tag the conversation. So Tom can use it as a follow up for his next meeting with that client. So he remembers, hey, what were the takeaways? Or it automatically puts those to do's inside of our ClickUp, you know, so it automatically is in our to do list. And then ultimately it's a search function. So it's like, hey, talk about phantom plans. And then you type in phantom plans and it pulls up like the last five conversations that have been about phantom plans. And then you can kind of scan that, figure out what it is. And so it's this like interactive wiki, that we're really that's like our main focus right now, I think before the end of the year is to really kind of lock that down and build that that part out. That's how we're leveraging AI now. But yeah. and just add to, I mean, what that could replace Adam, which I think is huge, is what our team does is often it's teams messages like, hey, does anyone have experience with phantom plans? But when people are dealing with so many emails and teams messages, those often get ignored.
Tom (00:18:29) - And that's not the best solution to do that. So if you did have a place where you're like, oh, I found all these conversations and in that you could say, oh, and Adam was in three of them. So if I feel like I need more, then I just go to Adam and say, hey, heard you talk about phantom plans with these three clients. Let's chat about my situation. Yeah, I think that's going to be huge.
Kyle (00:18:46) - Yeah, I love that that that use case. And you know, it's funny. Earlier in the year I said okay how do we make that happen. Right. How do we get this internal wiki, you know, to pull all of this information and you know, the immediate, you know, six months ago answer was, okay, well, you know, we probably need to hire a developer to, you know, put something like that together. Nope. Not anymore.
Adam (00:19:08) - Right.
Kyle (00:19:09) - Yeah, right. It's here. And that's kind of scary of how quickly that's come in to, you know, to give us those capabilities at a really reasonable price.
Kyle (00:19:18) - And so I love that idea of, you know, having it as an internal wiki, you know, even helping it feed into some of the deliverables you put out. Right. Like can it eventually, you know, create a memo, can it create a, you know, a reconciliation and you'll have to refine it as you go? But, yeah, I see a lot of power in taking the information that's coming in. From your calls, right? Your documentation and, you know, summarizing it and putting it into a great, great starting point.
Adam (00:19:47) - Yeah, there's a lot of learned information. The other one, the other big one, Tom, if you want to talk into it, it was like the code that they wrote to drop into Chat GPT for coding, right? Do you remember that? Like on our on our last team retreat, they basically they wrote like a big paragraph that you could take a client's entire financial statement, all their detail, like you take their GL, throw it in there and it would return back, like what accounts you should be coding everything to.
Adam (00:20:15) - And you type in like you just say, hey, this is client XYZ. And they're in, you know, in Dallas, Texas, and you drop in all the vendor names and it says, these are the accounts where everything should be coded and then give you the confidence level. This is a 95% confidence level, 50% confidence level, you know, let you know where all the codification should be for all your client data, which is pretty cool.
Tom (00:20:39) - Yep, yep. That was huge. Yes, I had forgotten that example. The other one that we had talked about was after stripping out any confidential information, could you feed in financial statements? And maybe I mean, one of my examples is so across the columns is all the different time periods and put that in. And I think you could both get back potentially accounting corrections like oh I see a big drop in accounting spend that may have you say do I need to do an accrual or something like that. But then the other side, what insights are you seeing.
Tom (00:21:06) - Might you come back and say, I see this trend going up, that maybe it either takes us a lot of time to find or we just don't even see that. But it's going through thousands of lines of data, pulling out stuff that we end up saying, okay, that really saves us time or gives us insights into that. Anything you're doing along that I love that the thought of insights, you know, kind of move from accounting transactions up to how do you get smarter doing it. Any other examples you think of Kyle in that area?
Kyle (00:21:32) - You know, I it's funny that you, you mention it because it's I was about to say the same thing of, you know, where with those insights. Right. All of a sudden, you know, a lot of people have been wondering, okay, how do I get into the Cass side of things? Right? How do I how do I become more of an advisor and less of an accountant? and, you know, insights like that, if it's spitting out places for you to look into, you know, that's such a great way for you just to immediately provide value, you know, to any of your clients.
Kyle (00:22:00) - Right? And all of a sudden it's, you know, hey, maybe we should look at this. You know, maybe I spend an hour or two to kind of dig into the details and say, okay, well, why is it lower? Right. And maybe I can, you know, give you a recommendation, you know, give a business owner a recommendation this month of, you know, we should do this or that. And I think those are the things that right now is everybody's looking for things like, you know, accounting advisors. Right, fractional CFOs, things like that. You know, that's where all of a sudden a lot of us that have a lot of, accounting and finance expertise, but not really sure how to use it in that, you know, advise capacity. it just aids that so much. So, you know, like you said, we're starting to get into some more of these engagements where, you know, we have where we basically have the, the setup of how to put together your monthly financial statements.
Kyle (00:22:47) - You're reporting package takes 15 minutes to put that together once it's set up. Right. you know, take and then you can spend the rest of your, you know, retainer or, you know, the hours that you spend digging into those insights. Right. And again, if to your point of if it's spitting out specific areas that you want to look into, that's that saves you even more time to focus on providing value to your clients. Right. And so, it's I think it's that way that you could really, you know, a lot of clients, a lot of firms could use that and automate that process. And all of a sudden you've got a pretty sustainable, process for, you know, even really high expertise areas like fractional CFO.
Tom (00:23:28) - Yeah. What about? We'll probably come back to some opportunities. But you want to talk. Should we talk just a little bit about some of the risks of doing it. And you've already mentioned one. I had kind of overreliance on technology.
Tom (00:23:40) - And when you talked about researching standards, you talked about citing your sources. So that could be one. But what other risks if someone's diving in, would you say maybe put the brakes on a little bit, or at least be really conscious of a certain area?
Kyle (00:23:51) - Yeah. So beyond, you know, validating the information you're getting, I'd say the other big one that maybe not a lot of people think of immediately is, what you're feeding into the system. And you made a good point of this earlier, Tom, of, you know, once you once you strip out, you know, the confidential information, then you can feed it in. I think there's a lot of, you know, people out there that don't think about taking that step and, you know, just may not be aware that you need to be taking that step. and that gets you into a lot of trouble if you're feeding confidential information into a large model like that. Right? I mean, I think we've already heard about a few, large companies where, you know, people were doing that.
Kyle (00:24:27) - Right. And now there's lawsuits, you know, coming out of it. And so especially as, you know, an accountant, a CPA, you have that really high level of confidentiality to your clients. you know, that's critical to be able to find a way to make sure that, you know, you're not dropping sensitive information, you know, into these models, right? Because it's going to get used and it's going to get, you know, shared with other areas. And so that would be the big that would be another probably key risk in my mind of whatever information you have, if it's, you know, a lease agreement, if it's the financial statements, right. It's, figuring out a good tool or a good way to strip out, you know, any of those names or information that that shouldn't be going in there?
Tom (00:25:11) - Yeah. And I I'll share a mistake I made early on. And so personally, I work with a not for profit.
Tom (00:25:17) - And they came to me in the IRS had given them a penalty. And so I used ChatGPT to write a abatement letter form. It did a phenomenal job. But in there I actually said, here's the organization, here's the time period, things like that. And then it was a little later to think about, like, you know what I need to do and I've done in the past is it's an anonymous company that you feed in and then you change it in the final draft. But it was amazing how it wrote the whole thing. But then there was sort of this aha! After like, okay, I just fed in the name and like what they did. So now that's gone into the internet and you could see easily making other mistakes like that where it's a lot more sensitive stuff and you're putting in stuff that it can use in the future. Yeah. For that. So I think that kind of clean up and what you're sharing ends up being a pretty important thing to be aware of what you're feeding in.
Kyle (00:25:58) - Yeah. And I think there are things coming into place that are going to help solve that problem too, right? I think, you know, with these things like custom GPT, like, you can have a dedicated instance, right? Have those walls up for your business, your organization. you know, where, you know, you can utilize, you know, maybe client names and, you know, it's not going out to the greater good. That's definitely something that you should be checking with, you know, your IP or your security, you know, legal counsel, you know, to make sure that you're not crossing any lines that shouldn't be crossed. but, you know, it's nice to see that they're starting to see where those risks are and they're trying to address them, you know, through future products. Models.
Tom (00:26:37) - Yeah and Adam mentioned one of the prevention ones when he talked about what we had done, the sort of confidence level that was in there.
Tom (00:26:44) - Adam, when you talked about the vendors, because it would be easy to say, oh, great, ChatGPT or whatever gave me all the coding within three seconds and just immediately apply all of it. And if you're not scanning through this confidential, you could probably make some really silly mistakes that you could catch by making sure you're doing some good review.
Kyle (00:27:02) - Yeah, absolutely.
Tom (00:27:04) - So, so maybe thinking of sort of going through and saying because it is just a machine doing it. And the same with like it doing its own coding and stuff like that, being careful that it didn't just pick up like a statement instead of a bill and go ahead and entered as a do bill and things like that. So that's probably one of the risks that you could run into is it can do something that used to take you an hour in like two minutes, but do you add some time to make sure there's a thorough review so you get better quality and not just say, great, I'll just pile tons more work on myself and everything runs faster.
Tom (00:27:31) - And suddenly I realize I've gotten out of control.
Kyle (00:27:35) - Yeah. Yeah. I would say make sure not to, you know, review, review, controls review is not going away as a process of this. Right. Like ultimately, we're getting to a more accurate place. But, you know, you should be treating it as you would for, you know, if a staff accountant is putting together, you know, a reconciliation, you know, doing that categorization, you know, you should be doing that same level of review and, you know, looking into things as you were before. The hope is that, you know, that any the errors will begin decreasing, you know, as, as the, as the model gets better and better. But yeah, just to kind of say, hey, because this is machine, you know, provided by a machine, you know, that's a dangerous road just to not do any type of review. So that's a really good point, Tom.
Tom (00:28:20) - I wonder about. You may not have a solution to this, but I think a challenge we have in the accounting industry, if you talk to people who are 20, 30 year people and they start out as auditors, often they'll describe the reason they know certain things so well is because even by hand or manually Excel, they had to repetitively do something that sort of drilled it into their mind. Now we've got people coming out that probably don't even have to do that. That's all being done by AI. And I wonder, do we run the risk of people not really understanding the basis because there was some value in doing some of those reconciliations to the point where you're like, I know this stuff in my sleep because it was so rude. Do you have any solution in your firm is you're like, okay, AI is doing the low level stuff, so you got to jump in knowing how to review.
Kyle (00:29:01) - Yeah, I'm thinking back to, you know, when I was an auditor and, you know, it seemed like where I learned most was when you were a senior.
Kyle (00:29:09) - Right. And all of a sudden, you know, you're not putting together, you're not the preparer of the workbook or the work paper or the reconciliation. Right. But you need to know it well enough to understand where there's issues. Right? Okay. This looks strange. And you have the bigger picture at the same time to say, okay, this is wrong. And this is why, you know, I think it's going to be important to keep that. I think that training is going to happen at the first level of review. Right? So, you know, if you're, an entry level accountant is no longer doing, you know, the preparation, right? If that's or the categorization, if that's happening from, you know, artificial intelligence, then their job role is probably going to be more of that first line of review. Right. Almost similar to if you're, right now. Right. I think a lot of firms are, you know, offshoring some of the lower level work to, you know, internationally.
Kyle (00:29:59) - Sure. but it is critical, you know, when it comes back to be looking through what happened, you know, and, and understanding, like, hey, this is, you know, we need to refine it this way. We need to look into this. We need to resolve that. I could I see there being a really good analogy between, you know, how that's happening, happening in offshoring right now versus, you know, I could see it being the exact same. So I would say it's making sure that that review and that training of that review is happening right then, because I think you get a lot of information on you know, why we're doing it, as well as how it's being done at that level. yeah. So that would be my thought, you know, right out of the gate.
Tom (00:30:38) - That makes a lot of sense. Yeah. Adam, I've heard you describe more on the offshore processes of you're going to give say me a task and here's what I want you to do.
Tom (00:30:47) - But then here's how I'm going to check that that work is done. And in some cases, you're even having me enter the information like go out and type in what the balance sheet says. And that number should match with the bottom of your spreadsheet says it'll show error. But you explain to me this is where I'm looking, so that when I get it and it doesn't have zero in there, I'm actually trying to fix it before I hand it to you, because I've already shown myself that it's wrong. I could see the same kind of thing with I if I'm going to feed it in. But here are the things I need to do after. So you're sort of building those controls in before you just jump in and try to do that.
Adam (00:31:18) - Yeah, it'd be great to be able to, you know, basically format or program in all your control totals and just be able to feed it straight through and have it come out and validate the work for you on the outset, so you know where to point to.
Tom (00:31:30) - Yeah, something a lot better than just here's your answer. And then you're like, okay, some black box. And I trust implicitly because it just seems so darn smart, right.
Kyle (00:31:38) - Yeah, I it, I and I really like, you know, back when you were auditing and you know we would review controls. Right. Both of you know how it's getting set up. And you always had to ask, you know, okay, what are the things that you're looking for in this work paper. And I think incorporating that more you know, not that everybody has to have socks going on, you know, for their review process. But you know, like you said, pointing out those key areas, control totals like if it's a specific area. Right. Oh here's the risk. You know, are you not supposed to add that? Are you not supposed to capitalize this? Right. you know, making that more specific. And I think you set that up. And as you refine it, that's not a huge task.
Kyle (00:32:15) - you know, after the initial setup.
Tom (00:32:17) - Yeah. I would guess lots of practitioners who are listening may hear a lot about AI and everyone sitting saying like, what would be a good first step? Or what should I be doing? Maybe I believe it, or maybe I don't, but I'm hearing enough about it. Any suggestions that you would have? And I guess I think of maybe some smaller firms who are like, we don't have a great tech department to suddenly help us do a huge initiative. What would you say are some good sort of initial use cases to help people step into this?
Kyle (00:32:46) - You mean like trainings or more.
Tom (00:32:49) - Training or first places to start? And I think of things like, while there are note takers like Adam mentioned, or you can go into ChatGPT and just start writing memos or things to say. Just touch the thing and see what it feels like. I'd just be curious kind of what you would say if someone says, I hear a lot about it and have no idea what to do.
Kyle (00:33:04) - Yeah, one place that we've started is, you know, when we with firms is, you know, any time, you know, a one of one of your teammates reaches out with a question, right. Okay. How do I account for this? What do I put this to? Like, what we say is okay, first put it into ChatGPT and see what it comes back with. And, you know, tell me what it comes back with. So we make sure that it's the right answer. But I think it starts getting that process going of, hey, you know, if I have a question, I'll enter it in, let's see how accurate it is. and when it comes back, you can see, okay, well, you know, is that feeding the right information? You know, am I getting confidence with the answers that are coming back, you know, from AI? so I would encourage, you know, to build that into kind of your staffs, you know, question process.
Kyle (00:33:52) - Right. Come back with a proposed solution. And, you know, that's what we're all looking for. You know, when we are reviewing and training our team members anyway. So, yeah, have them start using ChatGPT again, make sure there's no confidential information going into that. But, you know, and then and then you as a, as a firm owner say, okay. Yeah, this is helpful. Right? I'm getting, you know, fewer than 10 or 20 questions a day that I have to answer. And I think that's some low hanging fruit to save you time right out the gate.
Tom (00:34:21) - Okay, I really liked I hadn't thought about that one. Sounds a little bit like when a firm has the internal accounting expert like, hey, did you ask Kyle and did you run this past him? You're going to do the same with ChatGPT before I trust your answer. What did ChatGPT say about this? Yeah. What do you think about, do you have experience of meeting note takers? We've talked about that one just a little bit.
Tom (00:34:41) - Have you done much with that?
Kyle (00:34:42) - I haven't, I've seen a few, you know, note takers that have been pretty good. I know there's, one of our partners, diligence partners. You know, they use otter AI. which kind of does those transcripts? I know that, Microsoft is, you know, starting to roll out more of its copilot programs. You know, that focus on AI. and my understanding from that is even, you know, like you said, it's taking the information. It's not just, you know, giving you a transcript of the entire call, but like you said, you can ask it questions and it can fill in it can, you know, bring answers just based on what's in that call. so, you know, I think a lot of whatever systems we use on a daily basis, whether it's Microsoft, you know, whether it's Google, they're starting to build a lot of those tools in there.
Kyle (00:35:27) - but no, I think there's some really good providers, things like Otter AI, you know, that have been doing it for a while and they've gotten pretty refined at that. I don't know if you both have, other ones that you'd recommend.
Adam (00:35:40) - Yeah, I mean, we like I said, the teams went through about 15 of them. But I know the big one in the space right now is fireflies. that's the one that everybody seems to, to you hear the most about. And then, grain is another big one. And, and the one that we're ultimately landing on, like, is escaping me at the moment of alma. Of alma. There you go. Yep. There it is. so Tom knows. so we haven't officially rolled it out yet, but, there's just different aspects of it and controls that, that we feel like that one's ultimately the one that we're going to land with right now.
Tom (00:36:18) - Yeah. So and just the number of them. So Kyle, we are a remote firm where all of my client meetings are online. The number of people that are attaching those AI tools has just exploded in about the last six months. It's just been amazing to me, including Adam. I do a lot of coaching of firms, and we have every Tuesday fireside chat with people who have taken our virtual CFO course. There are 3 or 4 people who haven't been showing up to the meetings for months, but they send their AI to take notes. And so over and over, like, these people want to be admitted, and now we're up to like 2 or 3 every single week that they're just showing those things. But I'm most excited because I take a lot of notes during meetings. But I realize when I'm facilitating a meeting and trying to take notes, and then also usually I'm sharing my screen. I'm a terrible note taker. So I typed the whole meeting and then I go back to next week.
Tom (00:37:07) - And I almost always have like a sentence in like the most important part, like they made the decision to and then the sentence stops. And the idea of having I where I don't have to spend so much trying to type and think and talk at the same time is just great to say, okay, can it summarize what happened and maybe even come back and say, here's what you promised to do. And then, as Adam said, you can search across things too. Yeah. So I think that's going to be huge for us.
Kyle (00:37:28) - Absolutely. And I think you mentioned this a little bit earlier that, you know, if you miss a call, right, like you can go back and you can dissect, you know, you don't you have to listen to the entire 30 minutes to understand everything that was discussed. You can really get those key takeaways. I see that being huge for especially for internal calls. Right. So, you know, maybe I don't need to be on every single call.
Kyle (00:37:50) - But if I say, okay, you know, that was an important call, I need to know what came from it. You know, that saves you a ton of time to be able to focus, maybe even, like, have some focus time on a really important task. And, you know, later on you can kind of dissect, you know, what happens from some of these calls? I don't think it should be what you solely rely on. But I think if you're cutting it down, I think it's that that's a really good way to be more efficient with your time.
Tom (00:38:15) - Yeah, that's a really good point. One thing in Firefly's that Jamie is doing, and I hope Ifeoma can have this. So Jamie's our advisory director. If there was a point of an internal meeting he thought was really valuable, he would do just a video cut. And so he might say, hey, we had a great conversation about building your 2024 plans if something and watch this two minutes. Well, the chance of me watching a two minute video is pretty high, but the chance of me going back to watch the one hour video.
Tom (00:38:41) - I've got great intentions, but I don't do it. So I completely take your point that you could cut those things and then if you could search and find it yourself, but someone who says, hey, this was really valuable, it could be cool. It may also mean that two internal team people don't have to go to every client call where you could say, hey, this was the part that you really need to know because that topic did come up, but you didn't have to sit for an hour in the meeting just to make sure you were there for that one particular part of it. Yeah.
Kyle (00:39:05) - Absolutely. I'm sure clients will be happy, especially if anybody's, you know, getting billed on an hourly basis. Right. That's, you know, they're not paying for two people on the call versus one. Yeah.
Tom (00:39:16) - As we're winding toward the end. I'd love your reaction to this, and maybe you can put a time frame on this. I was at an AICPA meeting where the quote they said was that they don't think AI is going to be replacing accountants, except for accountants that completely ignore AI.
Tom (00:39:29) - And so I'd be interested kind of what your thought about that is. But maybe if you go out like a year or two, how big of an impact do you think we will see in that kind of time frame, for what's going to happen?
Kyle (00:39:41) - Yeah, yeah. I always go back to kind of the wave the wave analogy right of the, you know, there's going there's two types of people, right. There's those that are, you know, going to be really, you know, I want to say shouting against the wind, right. That say, hey, you know, this is awful. We shouldn't be using it. I'm not going to use it. you know, and there's those that are going to lean into it and say, hey, you know, I understand what's happening. You know, I'm going to try to use this for my benefit. Right? And so, you know, but that wave's coming. I think we can all, you know, see, just with how quickly things have been rolling forward, you know, even with people to say, hey, let's take a pause.
Kyle (00:40:16) - That hasn't happened. Right? So, you know, it's happening. It's coming quickly. And so, you know, if you are, I would say for anybody that's listening to this, right, it's if you are, you know, it's I'm seeing less of, you know, the benefit of sitting there and kind of shouting against, you know, artificial intelligence, it's happening, you know, the ones that really say, okay, it's here and it's staying, you know, how do I lean into this? How do I just use it for one simple task and start building on top of that? You know, I don't necessarily think that, you know, I was going to replace the accountants, you know, within the next year or two. But, you know, the accountants that use AI are going to, you know, really succeed. I would say, you know, versus others that kind of stick with the status quo.
Kyle (00:41:03) - And because, you know, they're not the amount of time that it's going to take to complete a task is going to go down considerably for a lot of these things, you know, versus firms that are kind of sticking with the status quo of, hey, let's get someone to, you know, spend eight hours on this. And I don't think that, you know, that that's a very sustainable, you know, business model. Yeah.
Tom (00:41:26) - Yeah.
Tom (00:41:30) - Yeah, I think we go out a year or so, just the. We had a conversation in May. We were in Arizona for our team retreat, and ChatGPT was a big topic. And I think what people are talking about is if we went back, I think we thought only 4 or 5 months, many of us had not even heard ChatGPT at the time, very little artificial. And all of a sudden a few months it was just everywhere. It seemed like. And I think that might be what we find a year from now is more the proliferation of things.
Tom (00:41:57) - As I mentioned, more people coming into meetings where maybe that just becomes the total standard or something, where I think my encouragement people would be dip your toe into, start using things like ChatGPT. And so just to start getting some experience, because I think if you're behind the curve and then begin, it's just going to keep going faster and faster where it's going to be pretty difficult to catch up.
Kyle (00:42:16) - Yeah, yeah. One forecast or thing that I'm curious to see, you know, in the coming years is, you know, they're talking more about, okay, well what's the next step here. Right. And it's probably, you know, an artificial intelligence, you know, app talking to another one. Right. For certain situations. That's that's a whole topic in itself. I won't go too deep down there. But, one thing that I'm interested to see is how, you know, audit firms start to use this, and interact with, you know, some of these, firms that are doing the preparation, right, or doing the consulting work and putting together things like reconciliation work papers.
Kyle (00:42:50) - Right. Is there a way that you can start, you know, building, you know, softwares or functionalities where, you know, they can communicate, they can be doing a lot of that? Okay, I'm checking the reconciliation and I'm spitting out the results to an auditor. And so they can go down and say, okay, well here are the focus areas that I want to look at. You know, I don't need to put together the the work paper myself. you know, if there's you're able to incorporate some automation into that as an audit firm. Right of. See. It's just seeing if it reduces just the overall back and forth when it's, you know, creating some type of, you know, work paper or otherwise.
Tom (00:43:27) - Okay. Yeah. That makes a lot of sense.
Kyle (00:43:29) - Yeah. I don't know how quickly that will come or who's going to put that together. But it's one thing that I'm very I could see that being an opportunity for someone. Yeah.
Tom (00:43:39) - Yeah.
Adam (00:43:40) - I don't know. I mean, yeah, the one thing that I think, I mean, my takeaway would just be that even for the people that are out there going, hey, I don't really use I. I mean, I think that in its purest form right now, I think a lot of people just do and they just don't realize it because it's built into a lot of the apps and the things that we're currently using right now. we didn't talk about anything revolutionary today that, like, is going to replace us. There's no robot that all of a sudden knows how to do all the accounting. I mean, don't get me wrong, it could change, you know, a couple of years from now, but nothing's really happening now. I guess my advice to people would be, I think the good news is, what ChatGPT did was open up the door for all these other platforms and softwares, to have this code to be able to build off of it. And even you don't have to be the most tech savvy in the world to know how to talk into something or get something to do something.
Adam (00:44:27) - You know what I mean? You know, you're not you're not having to be the person in there writing code. I think, you know, and I think that the people that are doing that great, knock yourself out. You're going to be doing a lot of stuff in ChatGPT have fun with that. But I think that, there's a lot of really smart people out there building software that are going to be able to do a lot of really cool things for us. And you're not really going to have to be that coder. So my advice would just be to, you know, have your ear to the ground, whenever it comes to exploring new software, make sure that you're, you know, doing things like the CPA where they talk a lot about this technology, find some forums. because again, I think it's less scary than what people think. I think that it's, I mean, right now people just feel like, you know, check http it's like, how do I code this or figure this out? Like I said, fireflies is pure AI, you know what I mean? And you don't have to do anything.
Adam (00:45:22) - You just log in and record a meeting and it does it for you. So, that'd be my only advice is just keep thinking about the technology and just make sure that you're staying focused on that because that's, you know, don't worry. There's going to be people that are going to build stuff for you. You don't necessarily have to build it yourself.
Tom (00:45:37) - Yeah. So here's my final question. And Kyle, I'll start with you and then Adam and me. So an example that you think is kind of fun from a personal standpoint from using something like ChatGPT or similar, anything you want to tell people like, hey, I either I do it this way, or I've heard of someone who did this and thought it was sort of a clever use of.
Kyle (00:45:54) - Yeah. Yeah. And I don't know, I feel like a lot of us may have used it this way so far, but, we're going to Costa Rica in 2024, right? So my wife and I want to take a trip out there and, you know, we don't know anything about Costa Rica, right? And so, you know, we said, all right, we you know, there's what are the options, right? You can go to a travel agent and say, hey, you know what? Can you help me book this? You can do a lot of research on your own and try to figure out, you know, what it looks like.
Kyle (00:46:21) - But my wife dropped that into ChatGPT and said, hey, we want to go to Costa Rica for seven days, like, give us an itinerary. We don't want to go to museums, right? Like, like you can add in those prompts and it you know, what it spit out was so helpful, so informative that it basically like ChatGPT is my new travel agent. So I love that idea.
Tom (00:46:43) - That's excellent. That's really cool.
Adam (00:46:46) - That is really cool because I get really tired of reading all the posts and then trying to figure out, like if it's just the, you know. So no, that would be really that'd be really super cool that that is a good, good idea. I actually, stole one from somebody that I heard from meal prep planning and stuff. They can, like, type in, like, all do all your ingredients in your cabinet and figure out what to make and give you different recipes to do and, like, build out a meal plan.
Adam (00:47:12) - Like, you can go in there and you can type in like what I like to eat, what I don't like to eat, and it'll build out your whole meal plan and help you even put together your store list and everything. So, tried that out and that's pretty cool.
Tom (00:47:12) - That's really good. The one in Adams heard this one before, but the, using it for gift ideas. Can you give me five gift ideas? And my daughter used this for a friend, and it was a friend of hers who lived in Cincinnati and was a Bengals fan and was turning 30 and 1 or 2 other things. And one of the suggestions was, why don't you give her a personalized Bengals jersey with her name on the back and the number 30 since that's her age, but she just she was using that and it gave her like ten ideas. And she just said it made me so much more creative to go through and just see these things and thought, I wouldn't think about that. And then she said, also sort of tested like, what are the different things you know about this person? And the better you can be, the better ideas that come out.
Tom (00:48:01) - So kind of fun things once you start using it. It's really fun to just keep putting in more and more things and just see what you can get out of it.
Kyle (00:48:07) - Yeah, those are some awesome ideas. I'm going to use both of those, that you guys recommended for sure.
Tom (00:48:13) - Good. Well, Kyle, this has been great. Thank you. I think hopefully informative for a lot of people I know it was for me and really enjoyed the discussion. Thank you very much.
Kyle (00:48:20) - Yeah, absolutely. Thanks for having me, guys.
Adam (00:48:23) - Yeah. Thanks.
Intro (00:48:23) - Enjoy this podcast? Visit our website summitCPA.Net to get more tips and strategy for achieving modern CPA firm success. We are here to be a resource in this ever changing industry.